model for pa 6.6

[Fabien]

hello,

I'd like to modelise failure of a solid composed of PA6.6 reinforced with glass fiber (with Castem).

Can I use the Von Mises strain as failure criterium?
Sould I take a failure criterium which depends on the hydrostatic pressure?

thank you for help,
good bye,
Fabien

[Jorgen]

:arrow: What type of loading is the part exposed to? Is it monotonic or cyclic?
:arrow: Does the part contain sharp corners or inhomogeneities that can cause high stress concentrations?
:arrow: Is the failure influenced by wear?
:arrow: What is the temperature and do you expect significant temperature variations or gradients?

The hydrostatic pressure can have an influence, and the Mises strain can be a reasonable failure condition to use. You might also want to use the chain stretch (http://www.polymerfem.com/modules.php?name=Downloads&d_op=getit&lid=35) failure condition.

Best of luck,
Jorgen

[Fabien]

Thank you for your answer :) ,

The part is exposed to a monotonic load of 5000N. This load should be applied 20 000 times.

The part is fixed using screws. There will likely be stress concentrations next to the screw holes.

The part is submit to outside climatic conditions.

I read the paper on UHMWPE. Do we have the same results for polyamides? How can I find the value for the limiting chain stretch?

The material I am willing to use is pa6.6 GF30 with characteristics :
tensile Modulus : 10000 / 7200 Mpa (dry/cond)
stress at break : 195 / 130 Mpa
strain at break : 3.3 / 5 %

I gess I should use the values given in the conditioned state, isn't it?

Here are two screen copies of the part :

http://www.crans.org/~cazes/1.png___http://www.crans.org/~cazes/2.png

good bye,
Fabien

[Jorgen]

Nice figures!

I recommend the following:
:arrow: Perform a finite element analysis of the Nylon 6,6 part when exposed to the maximum load (5000 N). This analysis will tell you how large the maximum stresses are in the part and how close you are to the monotonic failure stress and strain values.
:arrow: Find the fatigue behavior of the Nylon as a function of applied stress. Perhaps you can get this information from the raw material manufacturer, or you can use the handbook: "Fatigue and Tribological Properties of Plastics and Elastomers" by Plastics Design Library. This will tell you roughly how much the material strength is degraded when exposed to 20 000 load cycles.
:arrow: Apply a suitable factor of safety when evaluating the appropriateness of the design.

A few more comments:
:arrow: Note that PA 6,6 is sensitive to moisture and can loose about half its strength in a moist environment.
:arrow: The chain stretch failure condition is likely more accurate than the uniaxial stress or strain at failure. However, the only way to determine the critical chain stretch for your material is through a few simple experiments. Depending on how accurate you need the failure prediction to be, you may or may not want to go through that step.
:arrow: Yes, I would used the material parameters in the conditioned state.

- Jorgen

[hhspiny]

I would be very cautious when try to apply any polymer material model to fiber reinforced polymer in failure analysis (loaded significantly into nonlinear region). the presence of fiber completely changes the failure behavior and process, although initially, the stress/strain curve makes you think it is just another kind of polymer. Underneath, cracking starting from fiber tip, debonding etc starts to kick in very early, for PA can be as early as 1% of strain, and be failure process is controlled by microdamage evolution, not by the molecular chain deformation.

if the content of the fiber is large, say, over 20%. I would not use any polymer model, but rather use simple plasticity model calibrated by data, as it probably gives you as bad results as any polymer model. For more accurate/academic result, one needs to read into damage mechanics field. If the non-linear part is not of primary concern, only the failure point is, empirical laws such as deformation plasticity coupled with some failure cirteria, such as max principal strain, gives reasonable results.